Association studies of 1-butanol in n-heptane as a function of

Ronald E. Verrall, Diane K. Jacobson, Martin V. Hershberger, and Rufus W. Lumry. J. Chem. Eng. Data , 1979, 24 (4), pp 289–292. DOI: 10.1021/je60083a0...
0 downloads 0 Views 453KB Size
Journal of Chemical and Engineering Data, Vol. 24, No. 4, 1979 289

C M W

mol of H,O/mol of tert-butyl alcohol molecular weight weight percent

Subscripts 1 NaOH 2 water 3 tert-butyl alcohol

Llterature Cited (1) D. Sknonsen and E. R. Washburn, J. Am. Chem. Soc.,68,235-7 (1948). (2) G. L. Cunningham, U.S. Patent 2448868, 1948. (3) D. F. Othrner, R. F. White, and E.Trueger Ind. Eng. Chem.. 33, 1240-8 (1941).

Received for review July 14, 1978. Accepted June 25, 1979.

Association Studies of I-Butanol in n-Heptane as a Function of Temperature Ronald E. Verrail," Diane K. Jacobson, Martin V. Hershberger, and Rufus W.

Lumry

Department of Chemistry and Chemical Engineering, University of Saskatchewan, Saskatoon, Saskatchewan, Canada S7N OW0

The assoclatlon of 1-butanol In n-heptane has been lnvestlgated by near-Infrared spectrophotometry. Calculatlon of the average slre of 1-butanol polymers shows that no single 1-n polymer Is the predomlnant assoclated specles over the complete concentratlon range. The one parameter lnflnlte series model for alcohol assoclatlon is shown to be lnvalld for these alcohol solutions. The slmplest model which adequately represents the data In the dilute alcohol solutlons Is the 1-2-4. The unltary thermodynamic changes are calculated for thls model and are as follows: dimer AGO, (295 K) = -2.3 kcalldlmer, AHo = -4.3 kcalldlmer, ASo, = -6.8 glbbsldlmer; tetramer AGO, (295 K ) = -8.6 kcalltetramer, AHo = -12.9 kcalltetramer, ASo, = -14.6 glbbsltetramer. At thls level of aggregatlon stablllty Increases with degree of aggregation. There have been many reported studies' of self-association between simple alcohol molecules in nonpolar solvents. While it is generally agreed that there exist multiple equilibria among the many possible hydrogen bonded species, considerable controversy has arisen over the specific species present and their relative abundances. During the course of spectroscopic fluorescence investigations' of exciplex formation between indole and substituted indoles with l-butanol in n-heptane, it became necessary to obtain information on the degree of alcohol association in heptane and the concentrations of the lower-order associated species. This information was sought in order to interpret the mechanism of fluorescence emission from the excited indole compounds. A literature search indicated the only available information3 on the binary alcohol-n-heptane system was derived from studies of the fundamental OH vibrations. Data obtained from this region generally give considerable error in the derived results because of experimental complicationsand data analysis. As well, on the basis of criteria proposed by Smith? it appears that data from the fundamentals cannot yield high accuracy for association parameters at low concentrations. Luck5 also has pointed out that studies of the effect of concentration or temperature on the degree of association in alcohols are less subject to experimental error if derived from the overtone vibrations. For these reasons, studies were made of solutions of 1-butanol in n-heptane by using the infrared absorption of the first overtone of the 0-H stretch vibrations of the monomer. 002 1-95681791 1724-0289$0 1.0010

Experlmental Sectlon Solutions of 1-butanol (Fisher Spectranalyzed) in n-heptane (Fisher Spectranalyzed) were prepared by weight. Volumes were calculated by assuming zero volume change upon mixing and by using density data as a function of temperature for n-heptane and l - b u t a n ~ l . ~ Concentrations *~ ranged from 0.005 to 1.10 mol L-' with the majority of measurements being made in the range of 0-0.30 M. Fresh materials were used without further purification. Near-infrared spectra were obtained with a Cary Model 14 spectrophotometer over the spectral region 1340-1 500 nm. Matched I-cm Infrasil cells (with tight-fitting Teflon stoppers) were used for the sample and reference, and the cells were always filled so as to minimize any head-space loss problems.' The cells were contained in thermostatable cell jackets connected to an external constant-temperature bath (fO. 1 OC). Periodic checks of the temperature within the cell with an accurate thermometer showed no heat drift effects. The reference beam always passed through a I-cm cell filled with n-heptane maintained at the same temperature as the sample. Measurements were made at 6.0, 22.0, 31-0,and 44.5 OC so as to provide data as close as possible to the temperature conditions used in the fluorescence studies., As a check on the instrument and techniques, data were obtained independently by two operators at a single temperature, using fresh chemicals from the same supplier and a second spectrophotometer. The agreement between the results was within the errors reported below. By use of a double-beam technique with n-heptane in the reference cell, virtually all of any absorbance due to CH, groups of the hydrocarbon in the spectral region of interest is cancelled. Results In the analysis of the data it is assumed that the absorbance peak at 1405 nm is due solely to the monomer. Pimentel and McClelhn'a have suggested that the absorbance associated with the monomer cannot be related directly to the concentration of the monomer if linear polymeric species are present (-OH end groups of linear polymers may absorb at or near the wavelength of the monomer). Recent work,g however, has shown that the end-OH of linear self-association polymers do not contribute significant absorbance at the monomer peak in the fist overtone region so the analysis was carried out assuming the observation to be correct.

0 1979 American Chemical Society

290 Journal of Chemical and Engineering Data, Vol. 24, No. 4, 1979

Table I. Values of the Molar Absorptivity, Monomer as a Function of Temperature temp, "C

E,,

*

6.0 0.1 22.0 f 0.1 31.0 f 0.1 44.5 f 0.1

cm2mol-' 2.10 2.06 2.09 1.99

Standard deviation. used in extrapolation.

E,,

for the Alcohol

&

2.50

Nb

0.048 0.018 0.043 0.019

9 8

11 10

2.00

Number of low concentration points

The monomer overtone band of the alcohol showed structure, a weaker absorbance peak occurring at 1410 nm. The relative ratio of the absorbances of the two peaks in the monomer band remained constant with increasing alcohol concentration. These results appear to be similar to those observed" for alcohols in CCI, where the band structure has been shown to correspond to different conformations based on carbon-hydrogen substitution at the C atom adjacent to the OH group. However, at high degrees of aggregation packing should restrict conformational freedom and may be manifested in the intensity of this band. Absolute absorbances were measured relative to the absorbance baseline obtained with n-heptane in both sample and reference cells at each temperature. The latter were routinely determined before, during, and after each series of alcohol measurements at each temperature. The absorbance at 1405 nm, A, is assumed to measure the monomer at concentration C1 (vide supra) where the formal alcohol concentration is Co. The molar absorptivity, el, was calculated for the alcohol monomer by using two procedures. In the first method, from the relationship tcCo= A, a plot of A/Co vs. Co gives a constant value tl, as C - 0. A more :t O since accurate value can be obtained from a plot of t, vs. C it allows a more exact extrapolation of A typical plot for the latter is shown in Figure 1. While there is reasonable agreement between the values derived from both methods, only values obtained from the more exact graphical method are shown in Table I. These were obtained by fitting a first-order regression line to the data by use of a computerized leastsquares fitting program with weighting factors dependent upon the experimental errors in absorbance. The variation of t1with temperature is well within the error limits at all but the highest temperature. Even at that temperature it is not clear whether or not the observed decrease is significant and, if so,whether real or an artifact of the experimental method and data used to derive concentration values. Using the values of t1 reported in Table I, we can calculate the monomer concentration C1 in any sample from the relationship C1= Ale1. The total associated species are given by (C, - C,) in formal concentration. The fraction of monomer cy can be calculated from the expression cy = C1/Co. These values were fii to a power series (eq 1) in formal concentration

1.50

-

1.00

-

EC

0.50 -

=o Flgure 1. Determinationof

at low concentrations.

To gain some estimate of the degree of aggregation over the complete concentration range, we carried out the following analysis. I t has been previously shown by Hoffmann" that molecular weight data for alcohol solutions can be obtained by numerical integration of an appropriate function involving only monomer and formal concentrations. In general the following relationship exists between the formal alcohol concentration Co and apparent alcohol concentration CA, Le. ~ C =A(Co/Ci) dCi

(2)

The apparent alcohol concentration CAcan be obtained from the integrated form of eq 2 without reference to any particular association model by numerically integrating the curve Co/C1 vs. c1. CA =

derived value of q; the coefficients are reported in Table 11. Note that the monomer is the only species determinable with confidence from spectrophotometric data. Anharmonicity effects, intensity shifts in the fundamentals due to rehybridization and other specific details of the aggregates make the relationship

6.0 22.0 31.0 44.5

= f(t:C,)

Analysls of Data

using a least-squares program with weighting factors dependent

~~

with t,

between spectra and their concentrations highly uncertain.

on the errors in the experimentally determined absorbance and

Table 11. Coefficients of the Regression Equation CY = b , temp, "C b, bz

t,

Lc'(cO/cl)dC1

(3)

Values of CA were obtained by using a simple trapezoidal method for the integration. The average size of alcohol polymer can be definedldas

n = (CO - cl)/(cA - cl)

(4)

+ b2Cn+ b,C,' + . . . + b,CO4

~~

0.999 818 1.000 07 1.000 4 0 1.000 04

Average relative standard deviation.

-7.726 -5.097 -4.401 -2.488

97 65 04 62

b, 25.466 12.721 10.687 1.506

bs

b4

2 8 2 13

-33.823 -14.313 -13.843 1.959

3 2

1 75

15.062 5.748 8.943 -1.719

UQ

6 65 98 26

1.343 0.381 0.900 0.499

72 295 430 272

Journal of Chemical and Engineering Data, Voi. 24, No. 4, 1979 201 Table 111. Root-Mean-Square Deviations and Equilibrium Constants for Various Fits of I-Butanol-n-Heptane Near-Infrared Data model

T,"C

rmsd? M X lo4

1-2-w

6 22 31 44.5 6 22 31 44.5 6 22 31 44.5 6 22 31 44.5

1306 2550 1754 1862 251 56.9 25.9 8.8 333 117 68.1 17.3 250 26.0 5.5 1.4

1-2--

1-2-3

1-2-4

K,,,M-' 3.92 f 0.07 2.56 f 0.06 2.19 A 0.06 1.24 ~ 0 . 0 4 3.92 f 0.06 2.56 f 0.06 2.19 f 0.05 1.24 A 0.05 3.87 c 0.3 2.52 A 0.03 2.16 f 0.03 1.23 i 0.002 4.10 c 0.03 2.65 i 0.03 2.25 f 0.03 1.29 f 0.01

K,,, M-'

K,,,

K,, M-'

comments

9.98 f 0.12 5.88 i 0.10 4.85 i 0.14 3.27 A 0.09 9.63 f 0.10 5.44 i 0.07 4.54 f 0.12 3.05 i 0.11

analysis of concn range 0-1.0 M

analysis of concn range 0-0.15 M

analysis of concn range 0-0.15 M

38.1 f 3.2 38.1 i 0.4 10.3 i 1.6 4.67 A 0.13 1350 376 210 59

analysis of concn range 0-0.15 M

100 18 i 18 f 5 f A

a rmsd is the square root of the summed squares of deviations between C,(exptl) and C,(calcd) divided by the degrees of freedom (Le., number of data points minus number of parameters) in the system.

b. -1 L

..

' 0

.

P

4

.

-

2c1 -@

05

1

i

IO

12

C, ( M )

Flgure 2. Average size of 1-butanol polymers as a function of 1-butanol molarity in nheptane at four temperatures: (0)8 OC; (A)22 O C ; (0) 31 O C ; (A)44.5 OC. Note: not all of the low concentration data are

shown.

Figure 2 shows values of n obtained from the data as a function of formal butanol concentration. The magnitude of n should be invariant as a function of concentration within experimental error if only a single polymer of order n is formed. I t is clear from this figure that no 1-n model is a valid approximation for treating 1-butanol association in n-heptane. This is in agreement with previous resultsld obtained for the ethanol-n-hexadecane system. The polymer size is close to 2 at very low 1-butanol concentrations and increases to values of n 2 4 at the highest formal alcohol concentrations. There is scatter at higher concentrations and this suggests that the data are not as accurate as in the lower region. Further corroboration of the above observations was obtained in the following manner. If the presence of only one polymer of order n is assumed, then one can write c', = C1

+ nK,,C,"

and log (C,

-

c2

c1

+ c2 *

c3

(7)

are considered. In order to handle thls model one Is forced to make some simpllfylng assumptions. Kempter and Mecke12 assumed a slngle equilibrlum assoclation constant, K,, for addition of a monomer unit to any alcohol species. This glves the following expression for the formal alcohol concentration I

OO

concentration range. In such a model simultaneous equations of the type

0

C1) = log (nK,,)

+ nlog

c1

From eq 6 it can be seen that a plot of log (Co- C,) vs. log C1 should glve a slope of n if only one associated species is present. The plots obtained are nonlinear at all temperatures, with slopes of 2.1 f 0.1 and 4.3 f 0.4 in the low and high concentration regions, respectively. This is good evidence that at least two associated species are present. The data do not justify testing a wide variety of models. Several were used to test if a sequential indefinite self-association would describe the observed association over a wide

c, =

C1/(1

-

K,C#

(8)

Equation 8 can be rearranged to (C,

-

C1)/COC1

= 2K, - K,2C1

(9)

and a plot of the left-hand side of eq 9 vs. C1 should give a Such a graph straight line with slope Km2and intercept 2K,. of our data does not give a stralght line. Furthermore the slope is positive which predicts a physically impossible negative value of Km2.Clearly this model is not suitable and supports previously reported results on other alcohol-n-alkane systems. Another variant of the indefinite self-assoclation model assumes all associating species are present but that the molar equilibrlum association constant for dimer, K12,is different from all the other molar equilibrium constants; i.e., K12# Km = KU = ... = K., I t has been shownld that the expression for Co is given by

To evaluate K, from thls equation one must first evaluate K12. I t has been shown13 that K12may be obtained by plotting 2a2Co/(1 - a)vs. Co and extrapolatingto zero concentration. Values of the constants K12and K, are shown in Table 111. This model is better than the 1-m model but gives a rather poor representation of calculated Co values as judged by the magnitude of the root-mean-squaredeviation (rmsd) Table 111. Since this work was carried out specifically to provide Information necessary for interpreting photochemical processes at lower alcohol concentrations, fewer experimental results obtained in the high concentratlon range have undoubtedly contributed to a greater uncertainty in the association data In this region. Because of this fact, it was fen preferable to restrict the analysis to the concentration range 0-0.15 M. The arguments against the 1-~0and 1-n models presented earlier still hold for these conditions. However, when one considers the 1-2-m model, it can be seen that the fit, as judged by the magnitude of the rmsd value Table 111, is considerably

292 Journal of Chemical and Engineering Data, Vol. 24, No. 4, 1979

Table IV. AH"/Bond Values for 1-2-4 M o d e l

Table V. Thermodynamic Changes on Association

-AH/bond,a kcal mol-'

1-2-4 Analysis ~~

linear cyclic Total n-mer.

AHo,, 4.40 t 0.3 2.20 i 0.2

AHo,, 4.30 t 0.4 3.20 t 0.3

AGO,, (295 K) =-0.57'

AGO,, (295 K) =-3.48c

AH",,=-4.40a

AH",,=-12.9' Unitary Changes

AH" divided by n for cyclic n-mer and n-1 for linear

improved. Other association models such as 1-2-3 and 1-2-4 were also used to fit the data and as may be expected from the preliminary analysis of the data, the 1-2-4 model appeared to give the best fit over the lower concentration region (Table 111). In all cases the calculated K's show expected temperature dependence. The following points are worth noting concerning the derived results for the 1-2-4 model. Calculated values of K,, are approximately twice as large as that found149i5for alcohol dimer formation in other systems. However, as previously pointed out,ld when higher polymers are readily formed it is difficult to determine the value of K i 2 within f100% . As well, traces of nucleophilic contaminants introduce errors at the lowest alcohol concentration by overemphasizing the amount of aggregated materials, Co-C,. On the other hand, in agreement with other results," our results indicate that the dimer concentration in alcohol solutions is not small compared to slightly larger polymer species. The values of the constant K,, are of a similar magnitude to those reported by Tucker et al.ld for ethanoln-hexadecane using several models. The equilibrium constants from the 1-2-4 model were used to determine AHOIbond. The slope of the van't Hoff plot gives -AHoIRand the values of AH' association for both linear and cyclic polymers are shown in Table IV. The values are, however, 50 % smaller in magnitude than those previously reported3 for this system. The method of analysis does not allow a distinction to be made between cyclic and linear polymer species. I f one assumes the associated species are strictly linear then from Table I V it appears that there is no change in AHo per hydrogen bond with increasing number of hydrogen bonds. This is somewhat unusual as one would normally expect from water aggregation an enhanced stability with increasing number of hydrogen bonds. If, on the other hand, one were to assume that the polymers were all of a cyclic nature, then the hydrogen bonds show increasing stability with addition of n-mer. Resolution of the relative proportions of cyclic and linear polymer species requires other experimental data. For example, correlation of average polymer size with heat of dilution datal5 would at least provide limiting descriptions and independent estimates of the average hydrogen bond energy as a function of concentration. We would expect the aggregates to become more stable per mole of monomer as the aggregate site increases. In addition to the improvement in enthalpy of H-bond formation, the statistical factor based on the fact that there are two acceptor sites and only one donor H-bond site per hydroxyl group should increase with aggregate size. These factors are to some extent reflected in the list of unitary" changes (that part of the entropy and free energy change that is independent of the composition

-

AS",, =-32.0d

AS",,=-13.0b

(295 K), kcal AH",kcal AS",, gibbs AGO,

dimer

tetramer

-2.3 -4.3 -6.8

-8.6 -12.9 -14.6

a kcal/mol of dimer. gibbs/mol of dimer. tetramer. gibbs/mol of tetramer.

kcal/mol of

of the solution) given in Table V. Furthermore there are no factors which would tend toward condensation other than the rearrangements of H bonds possible in cyclic species and that is much limited by packing problems. Hence we suspect the unitary standard free energies to decrease by 2 to 3 kcal/mol of monomer up to somewhat higher degrees of aggregation. The equilibrium constants for higher aggregates can be estimated in a strictly phenomenological way by using the data of Table V. This procedure should be sufficient for most photochemical uses since alcohol concentrations are usually low. However, this approach obviously will not give much detail about the "inverted micelles" (large aggregates indicated as existing at high alcohol concentrations). With better data a cluster approach should provide extremely useful data comparing the self-association of alcohols (a deficiency in H-donor sites), hydrazine or hydrogen peroxide (deficiency in H-acceptor sites), and water (equal numbers of donor and acceptor sites). Literature Cited See, for example, (a) G. C. Plmentel and A. L. McClelbn, "The Hydrogen Bond",Freeman, San Francisco, 1980, Chapters 3 and 7; (b) A. N. Fletcher and C. A. Heller, J. Phys. Chem., 71, 3742 (1967); (c) A. N. Fletcher, ibM., 78, 2562 (1972); (d) E. E. Tucker and S. D. Christian, /bid., 81, 1295 (1977). M. V. Hershberger, R. W. Lumy, and R. E. Verrall, unpublished data, 1979. G. Geiseler, J. Fruwert, and H. Seael, Spectrochim. Acta, Part A , 27r, 1893 (1971). F. Smith, Aust. J. Chem., 30, 23 (1977). W. A. P. Luck in "Water A Comprehensive Treatise", Vol. 2, F. Franks, Ed., Plenum, Press, New York, 1973, Chapter 4. F. D. Rossini, Ed., "Selected Values of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds", Carnegie Press, Pinsburgh, Pa., 1953, p 287. R . C. Wilhoit and B. J. Zwollnskl, J . Phys. Chem. Ref. Data, 2, Table 59 (1973). C. A. Swenson, J . Phys. Chem., 71, 3108 (1967). A. N. Fletcher and C. A. Heller, J . Phys. Chem., 71, 3742 (1967). W. Diner and W. A. P. Luck, Ber. Bunsenges. Phys. Chem., 73, 526 (1969). E. G. Hoffmann, Z . Pbys. Cbem., Abt. B , 53, 179 (1943). H. Kempter and R. Mecke, 2.Phys. Chem., AM. B , 48, 229 (1940). N. D. Coggeshall and E. L. Saier, J. Am. Chem. Soc., 73, 5414 (1951). E. E. Tucker. S. B. Farnham. and S. D. Christian. J . Phvs. Chem.. 73. 3820 (1969): E. E. Tucker and E. D. Becker, J. Phys. Chem., 77, 1763 (1973). G. Brink and L.,,Glasser, J . Phys. Chem., 82, 1000 (1978). R . W. Gurney, Ionic Processes in Solution", McGraw-Hill, New York, 1953, Chapter 5.

Received for review December 18, 1978. Accepted July 12, 1979. This work was supported by the American Cancer Society (Grant No. BC-174).